token-wisdom

graph TD
    A[Prompt costs $0.02] --> B[Spend 2 hours optimizing it]
    B --> C[Save 500 tokens]
    C --> D[Now costs $0.015]
    D --> E{Was this worth it?}
    E -->|No| F[Learn when optimization matters]
    E -->|Actually yes| G[Prompt runs 10000x times]
    F --> H[Optimize the right things]
    G --> H
    H --> I[Save actual money]
    I --> J[Use savings to optimize more]
    J --> H

Optimization tips, tricks, and techniques. Also: knowing when optimization actually matters.

What goes here

• Token-saving patterns that don’t sacrifice quality
• Techniques for staying under context limits
• When to use which model/size
• Batching and caching strategies
• Measurements and benchmarks

The wisdom part

Not all optimization is worth it. Sometimes spending $0.50 in tokens is cheaper than spending an hour optimizing. This directory is about knowing the difference.